On Wednesday the Celtics traded Jaylen Brown — 2024 Finals MVP, five-time All-Star, fourth-leading scorer in the league this season — to the 76ers for Paul George, first-round picks in 2028 and 2031, and second-rounders in 2028 and 2030.
The reaction split exactly the way you'd expect. The scoring-title crowd says Boston gave away a 29-year-old star for an aging wing and lottery tickets. The analytics crowd — the people Brown recently dismissed as "AI hoops" — says Boston sold high on a name. One anonymous evaluator told ESPN they considered Brown "the seventh-best player on a team," which is the kind of quote that starts wars.
We run our own impact model. It watched every one of Brown's 5,840 possessions this season. So instead of arguing narratives, here are the numbers we publish every day on this site, and what they do and don't support.
The three numbers that matter
| Jaylen Brown | Paul George | |
|---|---|---|
| Age | 29 | 36 |
| PPG | 28.4 | 17.1 |
| TS% | 57.1% | 57.3% |
| Blended RAPM | -0.6 | -0.0 |
| O-RAPM / D-RAPM | -0.1 / -0.5 | -0.2 / +0.1 |
| Possessions | 5,840 | 2,664 |
Read that table twice, because both halves of it are true at once.
Brown scored 28.4 points a game — eleven more than George, on essentially identical efficiency. If you watch box scores, this trade is a robbery.
And: over 5,840 possessions — one of the largest samples in the league, so this is not a small-sample artifact — the Celtics were slightly worse per 100 possessions with Brown on the floor than you'd expect from the lineups around him. Not a disaster. Not "seventh-best player on a team." Just... roughly neutral, with the defensive number (-0.5) doing most of the damage. George, at 36, in an injury-shortened 49-game season, graded out at dead even.
Our model's view of the player swap, stated as plainly as we can: Philadelphia acquired eleven points a game of scoring volume; it did not acquire a measurable per-possession impact upgrade over the player it sent out. Boston converted a name into an equivalent-impact veteran plus four draft picks.
The trend is the uncomfortable part
RAPM on this site is published three ways: a prior (last season's information), a rolling number (recent play), and the blended value that feeds our game model.
Brown's arc this season: prior +0.6 → blended -0.6 → rolling -1.1.
That's the model changing its mind in one direction, all season, as evidence accumulated. The more it watched, the less it liked what the scoring was buying. His shot profile tells part of the story — an xPPS of 0.86 is well below league average shot quality, with only 17.7% of his makes assisted. He creates his own offense, which is genuinely hard, but it's hard offense, and the possessions data says it wasn't lifting the team's bottom line.
What the model can't see
We'd be doing the same thing we criticize if we stopped there. Three honest caveats:
The seventh-year age gap. RAPM measures the present. Brown is 29; George is 36 with a 49-game season behind him. Even if today's impact is a wash, the downside tails are not remotely symmetric. Philadelphia is betting on the age curve, and that's a rational bet a plus-minus snapshot can't refute.
Context is not destiny. Brown spent this season as a second option in an offense that didn't need his shot creation. Next to Embiid and Maxey he'll be asked for a different job. Players change value when their role changes — RAPM regularizes toward what happened, not what could happen. (The Sixers' front office is explicitly arguing Boston's lineup data understates him. That's not a crazy position; it's just not what the sample shows.)
Boston's return is the picks. If you grade this trade as Brown-for-George straight up, you're grading the wrong trade. Two firsts and two seconds from a team whose core will be 38 and 33 when those picks convey is real, possibly enormous, value.
The verdict, in our numbers
On our offseason tracker, where every official trade is graded by RAPM received minus RAPM sent, this one will land near zero on current impact for both sides the moment it's official — which is itself the analytics take: the most famous trade of the summer is, by on-court impact, a swap of roughly equivalent players, decided entirely by age curves and draft capital.
That's not "Jaylen Brown is bad." Nobody serious thinks that. It's that scoring volume and per-possession impact are different quantities, and this trade priced the first one.
You don't have to take our blend of the two on faith — the full splits are on the RAPM leaderboard, and you can put Brown and George side by side yourself. If the numbers change when Brown puts on a Sixers uniform, the same pages will say so. That's the deal with models: they don't have narratives, but they do have to show their work.
RAPM snapshot through April 18, 2026 — the end of the regular season. All figures are from the same data that prices our nightly game predictions.